S.Rahman/1

PROJECT FINAL REPORT

Project Title: Quantification of road dust and its effect on soil

PI: Dr. Shafiqur Rahman, Associate Professor, Agricultural and Biosystems Engineering,

North Dakota State University

Co-PI: Dr. Kris Ringwall, Director, DREC, North Dakota State University

Dr. Bernhardt Saini-Eidukat, Associate Professor, Geosciences, North Dakota State

University

Dr. Larry Cihacek, Associate Professor, Soil Science, North Dakota State University

November, 2016

S.Rahman/2

1) Project Summary:

Western North Dakota, USA is experiencing economic growth due to the rapid oil development. The

increased oil activities are also causing heavy vehicle traffic on the unpaved roads. Unpaved road traffic

may create coarse particulate matter (PM10) and fine particulate matter (PM2.5), and Total Suspended

Particles (TSP) emissions, and deposit them in adjacent areas. These PMs may affect human and animal

health, and soil quality. To address this issue, a study was conducted to characterize and quantify PM10,

PM2.5, and TSP generated from unpaved roads surrounding oil development areas. Particulate matter

concentrations were measured from two different sites using the miniVOL™ portable air samplers (Air

Metrics, Springfield, OR, USA) at three pre-selected distances from the road using the Federal Equivalent

Method. Additionally, composite soil samples were collected at the same location of PM sampling location

in one site (site 2) only. Scanning Electron Microscopy (SEM) was done on dust samples to determine

particle elemental compositions and X-Ray Diffraction (XRD) to identify minerals present. Inductively

Coupled Plasma - Mass Spectrometry (ICP-MS) was performed on soil samples taken from the same

location as dust samples to determine the elemental composition. The pooled average PM10 and PM2.5

concentrations were 30.84 ± 14.19 µg/m3 and 14.08 ± 6.56 µg/m3 from a periodically treated road (Site 1),

respectively; the pooled average PM10 and PM2.5 concentrations were 70.42 ± 38.37 µg/m3 and 19.60 ± 7.51

µg/m3 from an untreated loose gravel road (Site 2), respectively over a two-year sampling period.

Magnesium chloride was found to be the most effective treatment in reducing PM. Scanning Electron

Microscopy (SEM)/ Energy Dispersive Spectroscopy (EDS) analyses revealed that most of the particulates

were quartz (46%) or silicates (36%) minerals or biogenic particles (9%). Inductively Coupled Plasma –

Mass Spectrometry (ICP-MS) analyses on soil samples revealed that concentrations of most elements were

below the reference level measured by United States Geological Survey National Geochemical Survey

(USGS – NGS). This study improves our understanding of PMs in Western North Dakota, USA and

suggests avenues for future research to be taken for more in-depth study.

Overall goals and objectives

The overall goal of this project is to quantify road dust emission, especially PM2.5 and PM10 concentration,

in western North Dakota due to road traffic in the oil development area. Specific objectives are:

i) Quantification of PM2.5 and PM10 concentration,

ii) Quantify elemental composition of dust, and

iii) Quantify dust impacts on roadside soil quality and elemental composition of soil

2) Introduction and Rationale

Unpaved road traffic is a major source of dust nuisance and road traffic may emit considerable amounts of

coarse particulate matter (PM10) and fine particulate matter (PM2.5) emissions. Particulate matter has been

recognized as an air pollutant due to nuisance and its adverse impact on the environment, and can be a

pulmonary health risk (Mao et al., 2013; Pattey and Qiu, 2012). Fine particles (PM2.5) result from fuel

combustion from motor vehicles and power generation, while coarse particles (PM10) are generally emitted

due to vehicles traffic on unpaved roads, and materials handling, and as well as windblown dust. Inhalable

PM includes both fine and coarse particles. Small particulates (PM2.5) can be inhaled, resulting in

respiratory diseases and premature death (Gonzales et al., 2011; Pattey and Qiu, 2012; Samet and Krewski,

2007; Saxton et al., 1999; Donham and Thelin, 2006). Fine particulate matter on the road surface is also a

significant source of air pollution (Gunawardana et al., 2012). These particles can deposit in the respiratory

system and are associated with numerous health effects. Exposure to coarse particles is primarily associated

with the aggravation of respiratory conditions, such as asthma. In addition to health problems, PM is the

major cause of reduced visibility (USEPA, 2016). Increased dust in Western North Dakota, USA, from

unpaved road traffic is an inevitable consequence of increasing traffic from oil activities. The rapid growth

S.Rahman/3

of transportation activities is causing and releasing numerous pollutants to the environment and depositing

on the nearby road or crop land. Dust may contain heavy metals that may be toxic when their

concentrations exceeded certain thresholds (Guney et al., 2010). The amount of dust emissions from the

unpaved road is dependent on vehicle type, weight and speed of vehicle, wind speed and condition of the

road (Mao et al., 2013).

Road dust may affect plants, animals and humans that are exposed to it. Road dust is believed to affect both

the yield and/or marketability of crops grown alongside unpaved roads due to both physical and chemical

impacts. Dust can physically block stomata of plants and chemical characteristics of dust may affect either

soil or plants (Farmer, 1993). Dust cover on leaf surfaces may affect yield in a variety of ways, with the

yield reduction depending upon the thickness of cover and to an extent, the type of plant (McCrea, 1984).

The effect is likely to be greater on plants with young leaves as these retain a greater amount of dust, even

after a moderate rainfall. Similarly, dust may carry and cause plant disease and increased pest infestation.

Additionally, dust may also cause depressed appetite in livestock, which may result in a retarded growth

rate of around 20% for each day the animal is kept on the contaminated pasture (McCrea, 1984).

Agricultural worker exposure to dust or particulate matter is likely to result in mild to chronic respiratory

illness (Canadian Center for Occupational Health and Safety, 1999; Pattey and Qiu, 2012). It is reasonable

to postulate that oil field workers, truck drivers, and local residents similarly exposed would exhibit these

symptoms. Therefore, it is important to carefully quantify dust emission rates to assist in the development

of techniques or technologies to control dust from the source.

According to the US Energy Information Administration (USEIA, 2014), North Dakota's oil production

topped 1 million barrels per day in June, 2014, compared to 315,000 barrels per day only four years

previously in June, 2010. This significant increase in oil production has concomitantly increased oil rig

activities and traffic volumes in western North Dakota which is also causing noticeably increased dust

emission.

The chemistry of dust is important since it may contain a number of metals and elements which may be

concentrated in the smaller particles and they may travel further (Farmer, 1993). Studies suggest that heavy

metals may be concentrated within the first few meters of the roadside and their concentration may

decrease with distance from the road (Guney et al., 2010). However, their concentration in soil may

increase over time. Impacts of road dust due to oil exploration and production activities on surrounding

landscapes and vegetation are not well understood.

There is one recent study on dust control technology in North Dakota (Schwindt, 2013). However, that

study neither quantified the dust emissions nor measure the impact of dust on crop, soil, animal, and human

health. Prior to this study, no scientific data were available in western North Dakota on dust emissions and

their impact on soil health.

Therefore, there is an urgent need to quantify the dust emissions and adapt appropriate technology to

mitigate environmental impact. In this study diurnal dust deposition in a nearby livestock grazing area due

to traffic on an unpaved road was quantified and its impact on soil quality was monitored. This work

evaluated impacts of dust on landscapes surrounding oil field roads and/or drilling sites particularly with

regard to the distribution of potential elemental contaminants in dust distributed with distance across the

adjacent landscapes. The work evaluated any changes in the elemental composition of soil from the

beginning of the construction of the road out to areas up to 200 m from the road. In addition, changes in

soil organic matter, salinity and sodicity were evaluated.

Therefore, the objectives of this study were to i) quantify PM2.5 and PM10 concentration, ii) quantify

elemental composition of PM2.5 and PM10, and iii) quantify dust impacts on roadside soil quality and

elemental composition of soil.

S.Rahman/4

3) Method

Dust sampling

This study was conducted near the North Dakota State University Dickinson Research Extension Center

Manning Ranch Headquarters (Latitude: 47°12' N, Longitude: 102° 50' W), located in Dunn County about

35 kilometers north of Dickinson and 5 kilometers west of Manning, North Dakota, USA (Figure 1) (Site 1

and 2). Site 1, at the first location, was approximately 7 kilometers away from site 2. Monitoring sites were

established at DREC Manning Ranch, where one site is next to cropping trials and the another site is on

forage land. In both locations four EPA approved MiniVol portable air samplers (Airmetrics, Springfield,

OR, USA) were deployed. Samplers were swapped between sites. Initial sampling began with Total

Suspended Particles (TSP) to measure total dust accumulated on the filter in 24 hours. TSP contains

particulate matter up to a size of 45 micrometers (µm) in diameter. As the project moved forward, the TSP

sampling head was replaced with PM2.5 and PM10 sampling heads.

Fig. 1. Sampling locations at Manning ranch are shown with blue rectangular boxes. Site#1 has significant

well development; and Site#2, no well development yet

Sampling was carried out on the following dates (a table, or is this something for an appendix?)

Out of four MiniVol samplers, one sampler was installed upwind and three others were installed downwind

of the road sites depending on the prevailing wind directions at the monitoring sites to collect PM10 and

PM2.5 samples over 20 hour periods (Figure 2). Downwind samplers were installed at different distances

from the unpaved traffic to measure PM10 and PM2.5 deposition. The sampler was mounted to a pole

vertically as shown in Figure 3. After several samplings, the impactor was cleaned and greased for the next

sampling.

Before installing samplers in the field, 47 mm glass fiber filters or quartz filters were labeled with ultra-fine

Sharpie® pen while wearing latex gloves, stored in petri-dishes (48 mm in diameter), and conditioned in an

environmentally controlled room (relative humidity = 50.5 ± 0.2%, temperature = 22.6 ± 1.4 °C) at the

Nanoscale Science and Engineering (CNSE) Research lab at NDSU. Pre- and post-sampling weight of

Site #1

Site #2

S.Rahman/5

filters were taken by a microbalance (Sartorius CP2P) in an environmentally controlled room to determine

particulate matter mass with a resolution of 1 μg (0.001 mg).

Figure 2. Experimental setup of different locations: a. site #1; b. site #2; c. site #3

Figure 3. MiniVol™ Portable Air Sampler in operation at site #1 (vertically mounted)

To avoid static, a polonium bar at the back of the microbalance was used. During conditioning, filters were

reweighed 2-3 times for consistent weight. The differences in weight after each measurement should not

exceed 0.5% of the previous weight. After conditioning, filters were used within 7 days and protective

S.Rahman/6

holders were used during transportation. In the field, a cassette separator was used for insertion and

removal of these filters to and from the cassette. After inserting the filter, a pre-calculated actual air flow

rate was adjusted for the sampler and sampling was started.

During sampling, an air sample was drawn at 5 L/min, which is the recommended flow rate for PM2.5 and

PM10 using the MiniVol air sampler. A flow rate check was performed before and after each deployment of

samplers. All samples were protected in petrislides filter holders before and after sampling and weighed

twice in a 24-h period in order to obtain more accurate results.

Soil Sampling

Besides the dust sample, composite soil samples (a mixture of three samples) were collected at site 2 at

different distances (12 m, 30 m, 60 m, 90 m from the center of the road on both south and north side) using

a 25.4 mm soil core sampler (inside the cutting tip, diameter = 19 mm). Each soil core was taken at a 150

mm depth. Before collecting the soil sample in sampler bags, vegetation and ground litter were removed to

avoid contamination by plant materials. Then, the cores were stored in ziploc bags and labelled with a

unique ID. The sampling was done on a monthly basis to see the impact of dust on soil quality. These soil

samples were analyzed for potential changes is a soil elemental composition over time as well as for

potential chemical changes that could be used as “fingerprints” for evaluating dust issues in future studies.

Traffic Monitoring

Two battery-operated Simmons Whitetail Cameras (119234C) (Simmons Outdoor Products, Overland

Park, KS, USA) were used for tracking the number of vehicles passing through the study site (Figure 4).

Cameras were equipped with motion sensor and night vision (Infra-red) capabilities for capturing photos of

a passing vehicle even during night time. These cameras were set up on the same pole as the air sampler in

such a way that they would pick up both fast and slow moving objects. After each sampling, the number of

vehicles passed during the sampling period was counted and correlated with total dust emission during the

sampling.

Figure 4. Simmons Camera (deployed on site #2)

Meteorological Data

In-situ meteorological data (e.g., temperature, relative humidity, pressure, wind speed, gust speed, and wind

direction) were collected on site 1 using Hobo micro weather station data logger (H21-002) (Onset, Bourne,

MA, USA) (Figure 5). Additionally, weather data were also downloaded from North Dakota Agricultural

S.Rahman/7

Weather Network (NDAWN) and National Weather Service (NWS) to correlate important factors with the

dust emissions.

Figure 5. Onset Hobo Data Logger (H21-002) (Image partially from Onset Website –

www.onsetcomp.com)

Sample analysis

After sample collection, filters were transported back to NDSU and conditioned in the same pre-sampling

environment. During weighing and handling of filters, identical pre-sampling room conditions were

maintained and polonium sources were used to avoid any static during weighing of samples. To calculate

PM concentration for a sample taken with the MiniVol sampler, the volume of air that passed through the

filter at standard conditions, Vstd, or at ambient conditions, Vamb, was calculated as follows:

𝑃𝑀𝑎𝑐𝑡 = 𝑀𝑃𝑀

𝑉𝑎𝑐𝑡

Where: PMact = actual PM concentration, µg/m3 (actual condition); MPM = PM concentration, µg/m3

(Standard condition); Vact = Volume of air, m3 (actual condition)

The volume of actual air passed through the filter during sampling period at actual ambient condition

would be calculated as:

𝑉𝑎𝑐𝑡 = 60𝑚𝑖𝑛/ℎ𝑟 × 𝑄𝑎𝑐𝑡 × 𝑡ℎ𝑟

1000𝑙/𝑚3

Where: Qact = Flow rate of the sampler, liters/min; thr = Sampling period, hr

After weighing, particles from selected dust filters were analyzed for their physical and chemical

characteristics using Scanning Electron Microscopy (SEM), and Electron Dispersive Spectrometry (EDS)

(Pachauri et al., 2013; Tasic et al., 2006). Use of these complimentary methods was done to identify

particles (biogenic, geogenic, or anthropogenic), their grain sizes and size distribution, and their chemical

composition. SEM-EDS analysis was carried out for phase identification and semiquantitative chemical

characterization using a JEOL JSM-6490LV SEM at NDSU’s Electron Microscopy Center. Energy-

dispersive X-ray information was collected using a Nanotrace EDS detector with a NORVAR light-element

S.Rahman/8

window and Noran System Six imaging system (ThermoFisher Scientific, Madison WI, USA) at an

accelerating voltage of 15keV for the JSM-6490LV. In this analysis, samples were not coated with carbon

or gold because of the possibility of having biogenic organisms in the filters, because biogenic organisms

are basically made of carbon and oxygen (C + O > 75% of the total molecular weight), which may bias the

quantification (Figure 6). During EDS analysis, when the carbon was <10%, the carbon was excluded from

‘quant spectrum’ option. About 10-20 images were taken per sub-sample and when necessary, EDS was

done on the image by picking up several points. The magnification level used for taking the images was

x1500. The magnification level and spot size was fixed according to particles countered in the filters.

Figure 6. Sample preparation for SEM analysis: a. small sections cut from filter; b. sections placed on

carbon tape on cylindrical mounts

After collecting the soil cores from site 2, soil samples were air dried for at least 72 hours before

processing. The soil was hand crushed to pass through a 2 mm sieve and plant residues and rock fragments

were removed. A 10 g subsample of soil was taken from the bulk samples for ball-milling. The soil

subsamples were milled to pass through a No. 80 (80-mesh opening = 180 µm). Then, the prepared samples

were analyzed using Inductively Coupled Plasma Mass Spectrometry (ICP-MS) at Activation Laboratories

Ltd (Ancaster, ON, Canada) using the Ultratrace 2 method (aqua regia digest).

Identification of mineral phase from SEM results Relative atomic weight percentages were taken from SEM results to calculate approximate empirical

formulas. In the ideal case, each coefficient for a crystallographic site will be a whole number, and will

give the accurate mineral formula. SEM/EDS outputs data as percentage weight of atoms or oxides present

in the sample. These data together with the atomic mass of corresponding elements are used to calculate the

empirical formula of the mineral. The procedure is simple for pure phases and complex with phases with

impurities and trace elements. Figure 7 shows a sample SEM image and the EDS spectrum of quartz

mineral, and the calculation procedure for identifying the mineral.

S.Rahman/9

Elements O Si Na Notes

wt. % 53.06 46.40 0.54 Mineral: Quartz (SiO2);

Major elements: O, Si; Trace elements:

Na;

Nearly Spherical particles;

Formula: 2 O, 1 Si

grams/mol 16.0 28.09 22.98

mols 3.32 1.65 0.02

Normalized

mols

2 1 Negligible

Figure 7. Top: SEM image of filter showing particles, along with an EDS spectrum of a particle.

Bottom: Example calculation of possible mineral/phase group from SEM data (Quartz)

Reported weight percentages were divided by atomic mass to obtain the number of moles normalized. This

particular sample had the ratio of 1 mol Si and 2 mol O (and a trace amount of Na). So, it was determined

to be quartz. The procedure is similar for a complex mineral oxide weight as shown in Figure 8. In this

case, molar mass of oxide is used instead of the atomic mass. If the number of moles could not be brought

to a whole number, an approximate empirical formula is calculated. Morphology and knowledge in

mineralogy and crystallography can be beneficial in this case for identification. Figure 8 shows the

calculation for an aluminosilicate mineral phase. The mole numbers may not be brought to whole numbers

which could be due to the electron beam accuracy of the EDS method on small particles, and fluorescence

from silica filters.

Elements O Na Mg Al Si S K Ca Fe Notes

wt. % 47.58 1.18 0.61 15.63 26.55 0.70 3.01 1.42 3.31 Possible Mineral

Phase:

Aluminosilicates

Major elements:

O, Si, Al; Trace

elements: Na,

Mg, S, K, Ca,

Fe;

Nearly Spherical

particles;

Formula: 136 O,

2 Na, 1 Mg, 27

Al, 43 Si, 1 S, 4

K, 2 Ca, 3 Fe

grams/mol 16.0 22.98 24.3 26.98 28.09 32.06 39.09 40.07 55.86

mols 2.97 0.05 0.025 0.58 0.94 0.02 0.078 0.035 0.059

Normalized

mols

136 2 1 27 43 1 4 2 3

Figure 8. Calculating possible complex mineral formulas from SEM data (Aluminosilicates)

S.Rahman/10

RESULTS AND DISCUSSION

Particulate matter concentrations

Particulate matter (PM) is one of the Ambient Air Quality Standards (AAQSs) pollutants and

constitute a major class of air pollution (Cooper & Alley, 2002). Figure 9 shows the average PM

concentrations (PM10 and PM2.5) measured at site 1 over two years with respect to the number of

vehicles and the amount of rainfall during the corresponding sampling dates, while, Figure 10

shows the yearly average PM concentrations (PM10 and PM2.5). Table 1 shows the stepwise

regression analysis of PM at site 1.

Figure 9. Average PM concentrations with respect to traffic and rainfall at site 1.

0

2

4

6

8

10

12

14

16

18

200

50

100

150

200

250

201

5_

AP

R_

20

-22

201

5_

MA

Y_2

0-2

2

201

5_

JUN

_9-1

2

201

5_

JUN

_24

-26

201

5_

JUL

_1

3-1

5

201

5_

JUL

_2

8

201

5_

JUL

_2

9-3

1

201

6_

MA

Y_1

0

201

6_

MA

Y_1

1-1

3

201

6_

MA

Y_2

4-2

7

201

6_

JUN

_13

201

6_

JUN

_14

-16

201

6_

JUN

_28

-30

201

6_

JUL

_2

6-2

8

201

6_

AU

G_

03

-05

201

6_

AU

G_

15

-17

Rai

nfa

ll i

n m

m

Aver

age

PM

Co

nce

ntr

atio

n i

n µ

g/m

3

Sampling Date

PM10 PM_2.5 No. of Vehicles Rainfall (mm)

S.Rahman/11

Figure 10. Yearly average PM concentrations at site 1.

Table 1. Stepwise regression analysis results of PM concentrations at site 1

Particulate

matter type

Year of

sampling

Stepwise regression model equations

PM10

2015 & 2016

combined

No variables were statistically significant

2015 PM10 = 28.516 – 79. 55 × rainfall_in_mm

(r2 = 0.07)

2016 PM10 = 46.978 + 1.4615 × vehicle_count – 32.2796 ×

wind_speed – 0.37429 × wind_direction -51.06820 ×

rainfall_in_mm

(r2 = 0.30)

PM2.5

2015 & 2016

combined

PM2.5 = -0.75421 + 0.19184 × vehicle_count

(r2 = 0.16)

2015 PM2.5 = 0.98252 + 0.02480 × wind_direction + 63.485 ×

rainfall_in_mm

(r2 = 0.51)

2016 PM2.5 = -11.33165 + 0.43686 × vehicle_count

(r2 = 0.38)

Notes: - All variables here are significant at 0.15 level (p=0.15)

- r2 values reported are model r2 values

0.00

10.00

20.00

30.00

40.00

50.00

2015 2016

Aver

age

PM

co

nce

ntr

atio

n i

n µ

g/m

3

Sampling year

PM10 PM_2.5

S.Rahman/12

Based on 24-hour sampling period, the PM10 concentration value ranged from 7.55 µg/m3 to 60.5

µg/m3 in 2015 and from 0.79 µg/m3 to 261.45 µg/m3 in 2016. Similarly, the PM2.5 concentration

value ranged from 0.34 µg/m3 to 15.85 µg/m3 in 2015 and from 2.63 µg/m3 to 37.00 µg/m3 in

2016. Figure 10 shows the average PM10 concentrations in 2015 and 2016 were 25.33 ± 9.74

µg/m3 and 35.56 ± 20.25 µg/m3, respectively; and the average PM2.5 concentrations in 2015 and

2016 were 15.64 ± 6.65 µg/m3 and 20.25 ± 6.48 µg/m3, respectively. Most of the time, the PM10

concentration was below the NAAQS value (150 µg/m3). However, during June 28-30, 2016

sampling, the average PM10 concentration was 142.14 ± 89.28 µg/m3 which exceeded the NAAQS

threshold values. This may be attributed to the gravel road construction and ongoing well pad

construction near site 1 (15th St). The average PM concentrations for 2016 were slightly higher

than that of 2015 which might be due to drier weather and road construction/leveling activities. In

site 1, PM2.5 concentration value exceeded (37 µg/m3) NAAQS reference value for PM2.5 (35

µg/m3 for 24-hour sampling period) for one instance (June 28-30, 2016) during the sampling time.

Except for one or two incidents, the PM concentrations were below the NAAQS reference value

despite having high traffic.

PM emissions from a road likely depend on road conditions (dry vs. wet, treated vs. un-treated),

the number of vehicles, and weather conditions (precipitation, calm vs. windy), etc. The lower PM

concentration in site #1 during June 24-26, 2015 was likely due to road treatment. The road (15th

St) adjacent to site 1 was periodically treated with dust suppressants i.e., magnesium chloride. In

addition, there was 2.88 mm of rainfall during that sampling period, which might also contribute

to lower PM emissions. There was a significant drop in PM emissions during May 11-13, 2016,

which may be attributed to freezing and thawing effect as well as lower traffic activities.

A stepwise regression analysis was conducted to find out the impact of different variables on PM

emission. Statistical analysis revealed that, in 2015 sampling, the rainfall was poorly correlated

(r2=0.07) with PM10 concentrations; but, the combination of rainfall and wind direction had a

better relationship (r2=0.51) in the case of PM2.5 concentration emission at a significance level of

p=0.15. In 2016, PM10 concentrations were moderately correlated (r2 = 0.30) with vehicle passing

by, wind speed, wind direction, and rainfall. A similar or equally better correlation (r2 = 0.39) was

observed for the PM2.5 concentrations.

Figure 11 shows the average PM concentrations (PM10 and PM2.5) measured at site 2 with respect

to the number of vehicles and the amount of rainfall during the corresponding sampling dates.

Figure 12 shows the yearly average PM concentration at site 2. Table 2 shows the stepwise

regression analysis of PM at site 2.

S.Rahman/13

Figure 11. Average PM concentrations with respect to traffic and rainfall at site 2.

Figure 12. Yearly average PM concentrations in site 2.

0

2

4

6

8

10

12

14

16

18

200

50

100

150

200

250

201

5_

AP

R_

20

-22

201

5_

MA

Y_2

0-2

2

201

5_

JUN

_9-1

2

201

5_

JUN

_24

-26

201

5_

JUL

_1

3-1

5

201

5_

JUL

_2

8

201

5_

JUL

_2

9-3

1

201

6_

MA

Y_1

0

201

6_

MA

Y_1

1-1

3

201

6_

MA

Y_2

4-2

7

201

6_

JUN

_13

201

6_

JUN

_14

-16

201

6_

JUN

_28

-30

201

6_

JUL

_2

6-2

8

201

6_

AU

G_

03

-05

201

6_

AU

G_

15

-17

Rai

nfa

ll i

n m

m

Aver

age

PM

Co

nce

ntr

atio

n in

µg/m

3

Sampling Date

PM_10 PM_2.5 No. of Vehicles Rainfall (mm)

0.00

40.00

80.00

120.00

160.00

200.00

2015 2016

Aver

age

PM

co

nce

ntr

atio

ns

in µ

g/m

3

Sampling year

PM10 PM_2.5

S.Rahman/14

Table 2. Stepwise regression analysis results of PM concentrations at site 2

Particulate

matter type

Year of

sampling

Stepwise regression model equations

PM10

2015 & 2016

combined

PM10 =72.08904 – 51.12641 × rainfall_in_mm

(r2 = 0.07)

2015 PM10 = 99.74191 – 552.38530 × rainfall_in_mm

(r2 = 0.10)

2016 PM10 = 70.44775 – 2.63587 × wind_speed – 35.85478 ×

rainfall_in_mm

(r2 = 0.20)

PM2.5

2015 & 2016

combined

No variables were statistically significant

2015 PM2.5 = -14.91731 + 0.50329 × vehicle_count + 0.11719 ×

wind_direction

(r2 = 0.58)

2016 No variables were statistically significant

Notes: - All variables here are significant at 0.15 level (p=0.15)

- r2 values reported are model r2 values

The pooled average PM10 concentrations at site 2 were 70.42 ± 38.37 µg/m3 and PM2.5

concentrations were 19.60 ± 7.51 µg/m3 at standard pressure and temperature measured over a

two-year sampling period. However, the PM10 concentration over a 24-h sampling ranged from

1.91 µg/m3 to 253.60 µg/m3 in 2015 and from 5.17 µg/m3 to 179.66 µg/m3 in 2016, which is

higher than the pooled average concentration. Similarly, PM2.5 concentration value ranged from

2.56 µg/m3 to 52.91 µg/m3 in 2015 and from 3.27 µg/m3 to 31.52 µg/m3 in 2016. Figure 12 shows

that the average PM10 concentrations in 2015 and 2016 were 99.74 ± 57.88 µg/m3 and 45.29 ±

15.95 µg/m3, respectively, and the average PM2.5 concentrations in 2015 and 2016 were 25.94 ±

10.12 µg/m3 and 19.88 ± 5.27 µg/m3, respectively. Same as site #1, t the average PM10

concentrations in site #2 were lower in 2016 than 2015. This may be attributed to lower traffic

from decrease oil extraction activities in the sampling area.

In 2016, the average PM10 concentrations (199.32 ± 36.13 µg/m3) and PM2.5 concentrations (47.02

± 8.33 µg./m3) exceeded the NAAQS reference values (PM10 = 150 µg/m3; PM2.5 = 35 µg/m3 for

24-hour sampling period). This was likely attributed to various factors i.e., high traffic on a loose

gravel road and untreated road conditions. Also, traffic next to the sampling area likely

contributed to higher PM concentration, except May 20-22, 2015 when gravel was applied to the

road. For example, on May 20-22, 2015, the average PM10 concentration was 151.4 ± 58.32 µg/m3

that exceeded NAAQS PM10 value. The lower PM concentrations during June 24-26, 2015 was

likely due to lower traffic and 2.88 mm rainfall. During July 13-15, 2015, there were higher PM

concentrations compared to the amount of traffic which may be attributed to the ongoing

underground cable/pipe installation about 30 m to the south of the road, as well as drier road

S.Rahman/15

conditions. Despite of having comparatively higher traffic count, the PM concentrations during

May 11-13, 2016 were low likely due to the rainfall on May 10th (12.5 mm) and high wind speeds.

A similar pattern was also observed during May 24-27, 2016 (2.8 mm rainfall on May 25th), June

14-16, 2016 (8.4 mm rainfall on June 13th) and July 26-28, 2016 (2.88 mm rainfall).

From stepwise regression analysis, in 2015, the PM10 concentrations had weak correlation (r2 =

0.07) with rainfall but PM2.5 concentrations seemed to have a better relationship (r2 = 0.58) with

vehicle count and wind direction (p=0.15). In 2016, the PM10 concentrations had a weak

correlation (r2 = 0.20) with wind speed and rainfall, whereas PM2.5 had no association with other

factors (p=0.15).

Mineralogical characterization of particulate matter

Based on elemental composition and morphology, 299 particles were analyzed using SEM-EDS.

These particles were classified into three major groups: geogenic particles (derived from soil

sediments, weathered rock surfaces), anthropogenic particles (particles derived from industrial and

combustion activities), and biogenic particles (fungal hyphae with root outgrowth, organic plant

fragments, living micro-organisms). Some of them are explained below:

Geogenic particles

Most of the analyzed particles were found to be geogenic particles. These particles with crustal

origin include silicates of iron, magnesium, aluminum, calcium, quartz, Fe/Ti oxides, calcium

particles, chloride particles, carbonate minerals etc.

Quartz (SiO2) is one of the most common minerals found in on earth’s surface as it is a significant

component of many sedimentary, metamorphic, and igneous rocks. Quartz can occur in many

different colors, habits, and forms. Quartz crystals can be prismatic and can also appear in massive

form with no definable shape with no visible aggregate or crystals. The source of quartz can be of

natural origins. Quartz is characterized by high content of oxygen (O) and silicon (Si) (Si +

O>90% by weight) summing up to 100% with an atomic ratio of 1 Si to 2 O.

Non-quartz silicates particles are identified by high Si, aluminum (Al), O, and iron (Fe) content

with variable content of sodium (Na), magnesium (Mg), potassium (K), calcium (Ca), titanium

(Ti) with trace amounts of phosphorus (P) and sulfur (S), and sometimes carbon (C). Most

particles in this group showed irregular, sub-spherical, and spherical morphology. Possible

phases/minerals include feldspars, clays, oxides, carbonates, etc. Figure 13 shows possible

identification of aluminosilicates group (Al2SiO5) containing O, Si, and Al with lesser amounts of

Na, Mg, K, and Ca. It showed irregular morphology. Oxide minerals are identified by the high

content of O and other elements like Fe, Al with a low amount of trace elements. They tend to

have sub-spherical shape. Figure 14 shows an oxide mineral which has high Al and O content with

a trace amount of Na, Mg, K, Ca, and Fe. It likely is aluminum oxide.

S.Rahman/16

Weight O Na Mg Al Si K Ca Fe

158518(1)_pt2 50.27S 2.76 0.56 5.49 37.64 0.88 1.28 1.11

SEM ID Number Mineral

Group

Major Elements Minor Elements Morphology Formulas from SEM data

158518(1)_pt2 Silicates O, Si, Al Na, Mg, K, Ca,

Fe

Irregular 158 O, 6 Na, 1 Mg, 10 Al, 67 Si, 1 K,

2 Ca, 1 Fe

Figure 13. Particulate matter identification from actual samples (Silicate minerals -

aluminosilicates)

Weight % O Na Mg Al K Ca Fe

158428-A14(3)_pt1 40.50S 2.35 6.36 31.36 3.02 2.42 13.98

SEM ID Number Mineral

Group

Major Elements Minor Elements Morphology Formulas from SEM data

158428-A14(3)_pt1 Oxides O, Al Na, Mg, K, Ca,

Fe

Sub-spherical 42 O, 2 Na, 4 Mg, 19 Al, 1 K, 1 Ca, 4 Fe

Figure 14. Particulate matter identification from actual samples (Silicate minerals - oxides)

S.Rahman/17

Anthropogenic Particles

Anthropogenic particles include carbonaceous and industrial particles. Among industrial particles,

the dominant metalliferous particles contain Cr>40%, Mn>50% and Ni>10% by weight in

combination with trace particles. Very few industrial particles were found in this study as the

sampling sites were very far from industrial zones. Carbonaceous particles are significant as they

contribute highly to the total mass of the particles. In Figure 15, soot was identified. It had high

carbon content and low oxygen content. This type of particle can be produced from biomass and

biofuel burning. Earlier studies show that this spherical particle can scatter and absorb light (Cong

et al. 2008). Agricultural burning, tire residue, and waste incineration might be the origin of these

particles.

Weight % C O Na Mg Al Si K Ca Fe

158427-A14(2)_pt1 95.66 2.30S 0.13 1.92

SEM ID Number Mineral

Group

Major Elements Minor Elements Morphology Formulas from SEM data

158427-A14(2)_pt1 Soot C, O Al, Si Sub-spherical/

irregular

1653 O, 30 O, 1 Al, 14 Si

Figure 15. Particulate matter identification from actual samples (Anthropogenic minerals - soot)

Biogenic particles

Particles of biological origin were quantified by the method used by Matthias-Maser and Jaenicke

(1994). Both dead and alive biogenic aerosols contain minor amounts of Na, Mg, K, P, Si, Cl, Al

and Ca. These elements sum to approximately 10% of the whole weight of the particle. These

elements are also essential trace elements present in plants (Artaxo & Hansson, 1995). The rule to

identify such particles is: biological aerosols will have combined weight percentage of greater than

75% of carbon and oxygen, and phosphorus, potassium and chlorine will have weight percentage

of between 1% and 10% (Coz et al., 2010). S, Si, Zn, and Ca are also tracers of biogenic materials.

Figure 16 shows a round shaped outgrowth. It has a high carbon and oxygen content which sums

up to more than 75% by weight with trace amounts of K, Na, Cl, and Ca. The silica content is

probably from the filter fibers.

S.Rahman/18

Weight, % C O Na Si S Cl K Ca

158426-A10(1)_pt1 26.32 71.87S 0.16 1.33 0.10 0.12 0.06 0.04

SEM ID

Number

Mineral

Group

Major Elements Minor Elements Morphology Formulas from SEM data

158426-A10(1)_pt1 Biological

group

C, O Si, Na, S,

Cl, K, Ca

Round/spherical 2195 C, 4500 O, 7 Na, 47 Si, 3 S,

3 Cl, 2 K, 1 Ca

Figure 16. Particulate matter identification (Biological particles)

Such biological particles include microorganisms and fragments of all varieties of living matter

like viruses, bacteria, fungal growth, spores, pollen, plant debris, etc. (Cong et al., 2008; Coz et al.,

2010; Matthias-Maser & Jaenicke, 2000).

Elemental analysis of soil samples

The soil samples were analyzed to determine their elemental composition with regard to sixty

chemical elements by ICP-MS. Selected metals of interest were chosen because of their potential

impact on the local environment and, essentially, crops and human health. There were several

studies conducted in early 1980s which depict the elemental compositions of metals and their

reference values in the soil (Shacklette & Boerngen, 1984). In this study, concentrations of most

of the metals were lower than the reference values. Additional data collected by the United States

Geological Survey (USGS) Mineral Resources National Geochemical Survey (NGS) (USGS,

2003) (Sample ID: C – 250179, C – 237228, C-237239) and Smith et al. (2013) (Lab ID: C –

340224) were used to compare the measured values with these previously published values.

Evaluation of the analytical data from this study showed few elements that appeared to be

potentially influenced by dust from road traffic. However, some anomalies appeared to occur in

the data from mercury (Hg), lead (Pb), and nickel (Ni). Thus, these elements were examined more

closely. Figure 17 and 18 show the average mercury (Hg), and lead (Pb) concentrations with

increasing distances from the center of the road and their concentrations with respect to sampling

dates.

S.Rahman/19

(a)

(b)

Figure 17. (a) Average mercury (Hg) concentrations in ppm in soil at varying distances from the

road (n=8 at 12 m, n=6 at 30 m, 60 m, 90 m); (b) Average mercury (Hg) concentration in ppm

with respect to the date of trip (n=8 for all sampling dates except, n=6 for April 20-22, 2015).

The pooled (over a three years period) average Hg concentration was 0.046 ± 0.029 ppm. The

concentration of Hg varied from 0.020 to 0.10 ppm. The reference value from NGS was found to

be 0.020 ppm and 0.07 ppm (Smith et al., 2013). The average Hg concentrations on the north side

of the road were higher than that on the south side of the road. The mercury concentration is

highest in the month of May, 2015 and then decreased significantly. There is a rise in Hg

concentration in 2015, but the concentrations during 2014 and 2016 were low. It could be also due

to disturbance of soil on the south side of the road as the underground cable/pipe installation

activities were going on at that time.

(a)

(b)

Figure 18. (a) Average lead (Pb) concentrations in ppm in soil at varying distances from the road

(N=8 at 12 m, N=6 at 30 m, 60 m, 90 m); (b) Average lead (Pb) concentration in ppm with respect

to the date of trip (n=8 for all sampling dates except, n=6 for April 20-22, 2015).

The pooled average Pb concentration was 76.7 ± 168 ppm sampled over a period of three years.

The reason of large error range is due to high Pb concentrations on April 20-22, 2015 sampling

0.000

0.020

0.040

0.060

0.080

0.100

12 30 60 90

Aver

age

Hg c

on

cen

trat

ion, p

pm

Distance in meters

North Side South side Average

0.000

0.050

0.100

0.150

Aver

age

Hg c

on

cen

trat

ion, p

pm

Sampling date

0.0

100.0

200.0

300.0

400.0

500.0

600.0

700.0

800.0

12 30 60 90

Aver

age

Pb

con

cen

trat

ion

, p

pm

Distance in meters

North Side South side Average

0.0

100.0

200.0

300.0

400.0

500.0

Aver

age

Pb

con

cen

trat

ion

, p

pm

Sampling date

S.Rahman/20

date (222.3 ± 391.5 ppm), and on July 29-31, 2015 sampling date (97 ± 64.2 ppm). The Pb

concentrations varied from 6.4 to 435.9 ppm. The reference value from NGS was found to be 8.3

± 1.5 ppm and 15.2 ppm (Smith et al., 2013). The average Pb concentrations increased over

distance to the 60m sampling point and then dropped at 90 m. The Pb concentrations, like Hg,

were much higher in 2015 than that of 2014 and 2016’s. The high Pb concentrations during the

2015 sampling period could cause health issues to human working in the area and the animals that

are feeding off the site, if the soil is ingested. As the soil pH (5.92) was below 6.50, it could also

become available to plants.

Conclusions

The primary objective of this project was to quantify particulate matter (PM10, PM2.5) emissions

from unpaved roads (treated vs untreated) in well development area in the Western North Dakota.

Airmetrics miniVOL™ Tactical Air Samplers (Springfield, OR, USA) were used to quantify

PM10, PM2.5, TSP at selected locations. The pooled average PM10 and PM2.5 concentrations were

30.84 ± 14.19 µg/m3 and 14.08 ± 6.56 µg/m3 from a periodically treated road (Site 1), respectively

over a two-year sampling period. The PM10 emissions at site 1 were found to be weakly correlated

with rainfall in 2015 (r2 = 0.07) and moderately correlated with vehicle count, wind speed, wind

direction, and rainfall in 2016 (r2 = 0.30) at p=0.15. Likewise, the PM2.5 emissions were strongly

correlated with wind direction and rainfall in 2015 (r2 = 0.51) and with vehicle count in 2016 (r2 =

0.38). So, most of the time, the PM concentrations were high when the vehicle count was high and

PM concentrations were low when there was a rainfall event. However, the average PM

concentrations in 2015 were higher than 2016 but they were still below the NAAQS threshold

values. In addition, the PM concentrations were low when magnesium chloride was applied on the

road surface.

The pooled average PM10 and PM2.5 concentrations (over a two-year sampling period) were found

to be 70.42 ± 38.37 µg/m3 and 19.60 ± 7.51 µg/m3 from an untreated loose gravel road (Site 2),

respectively. There were some instances when the PM concentrations exceeded NAAQS values

which could be due to construction activities on road or high vehicle count or new gravel

application on the road or due to the untreated road surface. The PM10 concentrations were loosely

correlated (r2 = 0.10) with rainfall in 2015 and with wind speed and rainfall (r2 = 0.20) in 2016.

The PM2.5 concentrations were strongly correlated (r2 = 0.58) with vehicle count and wind

direction in 2015 and no correlation was found in 2016 (p=0.15). The PM concentrations in 2016

were lower than that of 2015 because of a decrease in oil rigging activities in the sampling area.

Elemental composition, and morphology of samples were analyzed by scanning electron

microscopy energy dispersive spectroscopy (SEM/EDS) which revealed there is a wide range of

minerals, biological aerosols, and little amount of anthropogenic particles in the area. 46% of the

particles analyzed were quartz, and 36% of the particles were found to be other types of silicates

which are basically constituents of road gravels in the sampling area. There were small amounts of

biological particles (9%), and oxides (7%). Very limited amount of anthropogenic particles (soot,

1%) was found in the area. The relative amount of quartz was higher at site 1 than that of site 2

which could be due to accelerated weathering process from a high number of traffic. Quartz and

oxides were predominant in PM2.5 samples too.

S.Rahman/21

Soil samples were analyzed using inductively coupled plasma – mass spectroscopy (ICP-MS) to

find out elemental compositions of metals present in the sampling area. To compare the measured

value, elemental compositions of metals from three reference sites were compared with the United

States Geological Survey (USGS) – National Geochemical Survey website and (Smith et al.,

2013). It was found that the concentrations of most metal decreased with increasing distances from

the center of the road to the north and south sides. Concentrations of the metals were higher in

some cases which were likely due to increased oil drilling activities, higher traffic, disturbance of

soil from underground cable installations, etc. The concentrations of most of the metals were

higher in 2015 than that of 2014 and 2016, during which, traffic and oil activities were the highest.

However, most of the metal concentrations were lower than the USGS reference values, thus road

dust may not pose any concern on soil quality based on this study. However, additional long term

studies are needed to evaluate the impact on soil and crop, as well as on human health and

livestock welfare.

Acknowledgment

We would like to acknowledge North Dakota State University Dust Research fund and as well as funding

from the Dickinson REC. Thanks to Sumon Datta and Dr. Md. Borhan who carried out sampling and

analysis. Special thanks to Dr. Kris Ringwall, Director DREC, for providing logistical support during

sample collection.

Products

Under this project, Mr. Sumon Datta has completed his MS degree in Agricultural and Biosystems

Engineering and title of his thesis is “Quantification and characterization of particulate matter generated

from unpaved roads in the oil development area of western North Dakota”. Sumon has also presented

preliminary research findings in the ASABE annual meeting (Datta, S., S. Rahman, M.S. Borhan, B. Saini-

Eidukat, and L. Cihacek. 2016. Quantification and characterization of particulate matter generated from

unpaved roads in the oil development area of western North Dakota. 2016 ASABE Annual International

Meeting, Orlando, Florida, July 17-20, 2016).

References

Cong, Z., S. Kang, S. Dong, X. Liu, & D.Qin. 2008. Elemental and individual particle analysis of

atmospheric aerosols from high Himalayas. Environmental Monitoring and Assessment,

160(1), 323-335. doi:10.1007/s10661-008-0698-3

Cooper, C. D. and F. C. Alley. 2002. Air pollution control: a design approach: Waveland Press. Coz, E., B. Artíñano, . L. M. Clark, M. Hernandez, A. L. Robinson, G. S. Casuccio, T. L. Lersch, and S. N.

Pandis. 2010. Characterization of fine primary biogenic organic aerosol in an urban area in the

Northeastern United States. Atmos. Environ. 44: 3952–3962.

Donham, K. J., and A. Thelin. 2006. Agricultural medicine: Occupational and environmental health for the

health professions. Victoria: Blackwell Publishing Asia.

Farmer, A. M. 1993. The Effects of Dust On Vegetation - A Review. Environmental Pollution 79(1):63-75.

Gee, G. W. and J. W. Bauder. 1996. Particle-size Analysis. In A. Klute (ed.) Methods of Soil Analysis,

Gonzales, H. B., R. G. Maghirang, J. D. Wilson, E. B. Razote, and L. Guo. 2011. Measuring cattle feedlot

dust using laser diffraction analysis. Transactions of the Asabe 54(6):2319-2327.

Gunawardana, C., A. Goonetilleke, P. Egodawatta, L. Dawes, and S. Kokot. 2012. Source characterisation

of road dust based on chemical and mineralogical composition. Chemosphere 87(2):163-170.

Guney, M., T. T. Onay, and N. K. Copty. 2010. Impact of overland traffic on heavy metal levels in

highway dust and soils of Istanbul, Turkey. Environmental Monitoring and Assessment 164(1-4):101-

110.

S.Rahman/22

Mao, Y., J. D. Wilson, and J. Kort. 2013. Effects of a shelterbelt on road dust dispersion. Atmospheric

Environment 79:590-598.

Matthias-Maser, S. and R. Jaenicke. 1994. Examination of atmospheric bioaerosol particles with radii > 0.2

mm. J. Aerosol Sci. 25: 1605–1613.

Matthias-Maser, S., and R. Jaenicke. 2000. The size distribution of primary biological aerosol

particles in the multiphase atmosphere. Aerobiologia, 16(2), 207-210.

doi:10.1023/a:1007607614544 McCrea, P. R. 1984. An assessment of the effects of road dust on agricultural production systems. Research

Report No. 156. Agricultural Economics Research Unit Lincoln College, Canterbury, New Zealand

Pattey, E., and G. W. Qiu. 2012. Trends in primary particulate matter emissions from Canadian agriculture.

Journal of the Air & Waste Management Association 62(7):737-747.

Samet, J., and D. Krewski. 2007. Health effects associated with exposure to ambient air pollution. Journal

of Toxicology and Environmental Health-Part a-Current Issues 70(3-4):227-242.

Saxton, K., D. Chandler, and W. Schillinger. 1999. Wind erosion and air quality research in the northwest

U.S. Columbia plateau: Organization and progress. In Sustaining the Global Farm: Selected Papers

from the 10th Intl. Soil Conservation Organization Meeting (ISCO '99), 766‐ 770. E. E. Stott, R. H.

Mohtar, and G. C. Steinhardt eds. West Lafayette, Ind.: USDAARS National Soil Erosion

Laboratory.

Schwindt, F. 2013. Investigation of methodologies to control dust on County roads in western North

Dakota. Report submitted to North Dakota Industrial Commission, January 2013. Contract no. G025-

054.

Smith, D. B., W. F. Cannon, L. G. Woodruff, F. Solano, J. E. Kilburn, and D. L. Fey. 2013.

Geochemical and Mineralogical Data for Soils of the Conterminous United States. In Sparks

et al. (eds.) Methods of Soil Analysis, Part 3. Chemical Methods. SSSA Book Series

5(3):575-1010. Sparks et al. (eds.) Methods of Soil Analysis, Part 3. Chemical Methods.

SSSA Book Series 5(3):961-601. Tasic, M., B. Duric-Stanojevic, S. Rajsic, Z. Mijic, and V. Novakovic. 2006. Physico-chemical

characterization of PM10 and PM2.5 in the Belgrade urban area. Acta Chimica Slovenica 53(3):401-

405.

USEIA, 2014, Crude Oil Production. http://www.eia.gov/dnav/pet/pet_crd_crpdn_adc_mbblpd_m.htm.

Accessed 07/07/2014.

USEPA, 2011. Particulate matter pollution from unpaved roads. www.epa.gov/oecaagct/trur.html.

Accessed 09/21/2011.